CZA Native Codec
The core engine analyzes byte streams, applies lossless compression suited to the data and restores the original down to the exact byte.
From reducing file size to protecting data
WITIVE Inc. has developed the core technology behind the internally designed CZA Native Codec and the CZA Archive Format, which safely stores files and directories. The new format uses the .cza extension and is planned for integration into WITIVE’s compression solution, CONZIP.
CZA is not a file format that merely bundles existing compression engines. Building on proven compression theory, WITIVE independently designed CZA’s token model, context construction, block-level decisions, bitstream and profiles, then developed them into a distinct specification and implementation.
CZA Archive Format
A new archive format that stores files, folders, metadata, indexes and integrity information in a single .cza file.
CONZIP
The product layer that connects CZA to an accessible compression interface, command-line tools and a future SDK.
Why the codec and archive were designed together
CZA Native Codec handles compression and decompression of pure data streams, while CZA Archive Format handles file lists, directories, metadata, indexes, path safety and storage procedures. Separating these responsibilities is intended to preserve restoration compatibility for existing files even as the compression algorithm improves.
The design analyzes each block and selects a suitable mode while managing container and codec bitstream versions independently. This decision accounts for both the long-term recoverability of stored data and future feature expansion.
Three compression profiles for different workloads
Fast prioritizes low latency and bounded memory, Balanced targets a balance between compression ratio and processing speed, and Maximum prioritizes compression ratio. Users can select a profile according to the job and system environment.
CZA aims for deterministic encoding that produces the same output from the same input and settings. Decoder limits on output size, memory, blocks and work prevent abnormal input from consuming excessive system resources.
Exact restoration and integrity come before compression ratio
The first criterion in CZA development is 100% lossless restoration. The system incorporates block- and frame-level error detection, archive metadata and payload verification, fail-closed rejection of truncated or oversized input, and extraction procedures that block unsafe paths.
Internal development validation confirmed matching SHA-256 values between test inputs and restored outputs, together with round-trip tests on Windows and Linux. In a limited 2 MiB C source-code test, CZA Fast produced smaller output than LZ4 and Snappy. WITIVE does not present that result as a generalized best-performance claim and will continue validation across broader data categories and external environments.
Starting with CONZIP and expanding across storage technology
WITIVE plans to connect CZA as a new compression option in CONZIP and refine compression ratio, speed, memory usage and compatibility with real files and environments through beta distribution.
The company then aims to expand the technology through CLI and SDK support, large-file streaming, split archives, local deduplication and cloud-storage integration, building a proprietary data-storage technology ecosystem.
Author's Note
